Certera.AI by Percipient is a comparison and evaluation platform for legal AI, designed to measure how large language models handle real legal work through blind, head-to-head testing.

Key features and functions include:

Blind Head-to-Head Model Comparison
Sends a legal prompt to two AI models and displays both answers side by side. Model identities remain hidden until after the user selects a preferred answer, so evaluation is based on answer quality rather than model reputation.

Model Selection and Direct Querying
Allows users to choose specific models to compare, or query an individual model directly, in addition to blind comparisons.

Document Attachment
Supports attaching documents to a query for review or redlining as part of a comparison.

Model Coverage
Includes AI models from major US labs, open-source developers, and Chinese labs, with 55 models available as of its beta launch.

Sensitive Information Anonymization
Detects and removes sensitive information, such as names and Social Security numbers, from submitted prompts.

Leaderboard
A lawyer-voted, Elo-based leaderboard ranking model performance is planned, along with context-specific rankings by practice area, such as contract law or litigation.